Knock counts. What's "bad"

[buddrow]

Ok so i have been slowly adding timing and keeping an eye on knock counts. Now im not getting large counts but i do usually get 5-10 during normal driving and hard acceleration goes as high as 45ish.

My question is this, on the 747 ecm, at what point does the ecm enable retard?

If i just ned to read thru TP i can when i get home but im feeling lazy tonight so if someone has a quick, or not so quick expaination please share.

thanks for reading,

Buddrow

[delcowizzid]

any amount of knock counts is bad.is it real knock ? things like bad engine mounts and exhaust banging on the body will trigger knock as well.so will roller rockers and big cams etc.

[RobertISaar]

My question is this, on the 747 ecm, at what point does the ecm enable retard?

assuming it acts like a P4, any/every detected knock event can cause retard. the way the knock accumulator is presented over the datastream, only the MSB gets transmitted. the LSB can/will increment without being able to see it over the datastream. generally, the LSB will NEVER increment one count at a time, the knock filter holds the circuit in "knock present" for too long for that to happen. so instead, the LSB can jump up 20+ counts for a single detected knock. when it rolls up to the 255 range, it will reset to 0 and the MSB will increment.

if you wanted to get VERY obsessive about it, you'll need to watch the LSB instead of the MSB.

one of the symptoms of only the MSB being sent over is that you'll occasionally see little bumps of knock retard(like 1* or so), but no increase in the knock counter. on a 160 baud unit, i don't think that would ever be visible unless you're using something like the EBL that has stupid-fast update rates.

[84Elky]

Expanding on what RIS said in Post #3, right now you are seeing in your logs one knock count reported for every 255 knocks actually detected by the sensor. If you want to see every knock actually detected, just open the bin editor within TPro and change the value at address 0x508 in your bin file from 0x73 to 0x74. Now every knock detected will be reported in the logs. When done, just change 0x74 back to 0x73, or leave it as 0x74. With 0x74 reported, you will see the actual knock counts the code used to invoke knock-related spark retard.

[damanx]

tag.

[EagleMark]

That's a lot on a 7747. Not sure but think it's quite different then a P4? Remember your only getting data every 1.2 seconds so this is an accumulation of what has happened. Then it stays at that number for all cells until it increases again.

To even find correct location in data log it takes watching the data one frame at a time while watching history table, then match it up to spark table and take some out.

If it's happening at WOT? Do the Highway Lean Cruise and ESC done patch, can then turn off HLC by rasing temp to max. This turns off the forced knock test at WOT...